Finds embedded URLs in a web page and remaps them to https scheme given a few limited logic rules

findAndRemapHttpURL.groovy

/*** BEGIN META {
  "name" : "findAndRemapURL",
  "comment" : "Finds embedded URLs in a web page and remaps them to https scheme given a few limited logic rules",
  "parameters" : [ 'vWebPageUrl','viiifCurrentPort','viiifRemapPort'],
  "core": "2.222.1",
  "authors" : [
    { name : "Ioannis K. Moutsatsos" }
  ]
} END META**/

//Note Includes special case for also remapping iiif http end points to a different https port

import java.util.regex.Matcher
import java.util.regex.Pattern
import hudson.model.*

jenkinsUri=new URI(jenkins.model.Jenkins.instance.getRootUrl())
jenkinsLocalHost=jenkinsUri.getHost()


//Pattern urlPattern = Pattern.compile("\\b(https?|ftp|file)://[-a-zA-Z0-9+&@#/%?=~_|!:,.;]*[-a-zA-Z0-9+&@#/%=~_|]",Pattern.CASE_INSENSITIVE);
Pattern urlPattern = Pattern.compile("\\b(https?|ftp|file)://[-a-zA-Z0-9+&@#/%?=~_|!:.;]*[-a-zA-Z0-9+&@#/%=~_|]",Pattern.CASE_INSENSITIVE);



artifactURL=vWebPageUrl
iiifCurrentPort=viiifCurrentPort
iifRemapPort=viiifRemapPort
  //example Jenkins URLs pointing to artifacts and reports with embedded http://myJenkinsServer.net:8080 style Jenkins references
  //"https://myJenkinsServer.net/job/IMAGELIST_GALLERIES/93/artifact/reports/Column-10_AND_RowNumber-4_AND_FieldIndex-4_AND_DNA-PROBE_Contribute_ImageList_748_20211012121922.html"
  //"https://myJenkinsServer.net/job/JOB_CONFIG_ANALYZER/875/artifact/buildReportData/paramProps.csv"
  //"https://myJenkinsServer.net/job/JOB_CONFIG_ANALYZER/875/console"
  //"https://myJenkinsServer.net.net/job/UTIL_CSVDATA_REVIEW/15/artifact/dataTableSkipRows.html"
def urlText=  new URL(artifactURL).getText()
def reMap=[:] //a list maintaining URI remappings


Matcher matcher = urlPattern.matcher(urlText);
while (matcher.find()) {
  urlMap=[scheme:"", host:"", port:"", path:"", query:null, fragment:null]
	String address = matcher.group()	
    URI uri = new URI(address);
 	urlMap.scheme=(uri.getScheme()=='http')?'https':uri.getScheme()
    urlMap.host=(uri.getHost()=='localhost')?jenkinsLocalHost:uri.getHost()  
    urlMap.port=(uri.getPort()==viiifCurrentPort as int)?viiifRemapPort as int:uri.getPort()
    urlMap.port=(uri.getPort()==8080)?-1:urlMap.port
    urlMap.path=uri.getPath()
//Remapping iiif end points it's a bit trickier see: https://iiif.io/api/image/3.0/#3-identifier  
  if (urlMap.path.startsWith('/iiif/2/')){
    println 'Got iiif EndPoint,port:'+uri.getPort()+'\r\t'+address
    urlMap.path='/iiif/2/'+ (address.split('/iiif/2/')[-1])
    remappedUri=new URI(urlMap.scheme+'://'+urlMap.host+':'+urlMap.port+urlMap.path)
    println 'New iiif EndPoint: '
    println '\t'+ remappedUri
    println '-'*10
  }else{  
    String domain = uri.getHost();
    println("Got URL,port:"+uri.getPort() +'\r\t'+ address)
    remappedUri=new URI(urlMap.scheme,null,urlMap.host,urlMap.port,urlMap.path,null,urlMap.fragment)
    println("New URL: ")
    println '\t'+remappedUri  
    println '-'*10
  }
	reMap.put(address, remappedUri.toString())
}


reMap.each{k,v->
  urlText=urlText.replace(k ,v)

}
 
return urlText

// References: 
// https://gist.githubusercontent.com/sagrawal31/55451ee85130f2dcda8e/raw/974800a6a52c2aea0d0521ece031d10a6950a1c3/DownloadURLs.groovy
// 1. http://stackoverflow.com/questions/5713558/detect-and-extract-url-from-a-string
// 2. http://stackoverflow.com/questions/4674995/groovy-download-image-from-url